ctp-signature

This repository - in a multi-center cohort (n = 131) - aims at predicting individual's pain sensitivity using grey matter cortical thickness (gmCT) measure. The repository contains the following:

• calculated values of gmCT using freesurfer >> see data folder,
• various potential confounders (phenotypes) from MRI, demography & psychometrics aspects >> see data folder,
• dedicated machine learning models based on LASSO linear regression method >> see notebook folder - also see figure below for a flowchart,
• results >> see output folder
• the methodology and the results have also been selected for a poster at the joint ISMRM-ESMRMB & ISMRT-2022 - London - UK and at the OHBM-2022 - Glasgow - UK (please see below).

Once the repository is downloaded with the essential programming packages, the obtained results can be re-produced using the jupyter notebooks. To redo the analysis by your own, which includes calculating the cortical thickness measures, please follow the hints below:

• use datalad to acquire the multi-center datasets(T1w - MRI images) using the commands below.
• Check out DataLad (https://handbook.datalad.org/en/latest/usecases/openneuro.html) for more info on how to use this download method.

RPN-signature_Study1:

datalad install https://github.com/OpenNeuroDatasets/ds002608.git

RPN-signature_Study2:

datalad install https://github.com/OpenNeuroDatasets/ds002609.git

RPN-signature-Study3: Data is available upon reasonable request.

Dockerized container

The docker image for the ctp-signature is avaialble at: https://hub.docker.com/r/pnilab/ctp-signature.

• To get started, one needs a freesurfer license.txt, that can be obtained from: https://surfer.nmr.mgh.harvard.edu/registration.html
• pull the docker image:

  docker pull pnilab/ctp-signature:latest
  

• use the following command to estimate cortical thickness, and for calculating ctp-score:

  sudo docker run -ti --rm \
  -v <path to bids parent directory where datasetsats present>:/bids_dataset \
  -v <path tp bids derivatives directory to store all output>:/output \
  -v <path of freesurfer license.txt>:/license.txt \
  pnilab/ctp-signature:latest /bids_dataset /output participant --participant_label 0001 \
  --license_file "/license.txt" --ctp 'true' --skip_bids_validator
  

Note

• if you do not want to use --skip_bids_validator, then make sure to install the entire datasets using datalad, and not just the reference links.
• a good practice is to use the same name of directories inside the docker while mounting and for code minimization. An example of the above code itself is shown below:

  sudo docker run -ti --rm \
  -v /BIDS_dataset_directory:/BIDS_dataset_directory \
  -v /freesurfer_license_folder:/freesurfer_license_folder \ 
  pnilab/ctp-signature:latest /BIDS_dataset_directory /BIDS_dataset_directory/derivatives participant --participant_label 0001 \
  --license_file "/freesurfer_license_folder/actual_license.txt" --ctp 'true' --skip_bids_validator
  

ML-pipeline

OHBM-2022 poster presentation